Merge pull request #711 from opencb/TASK-6780

TASK-6780 - Port Patch 6.2.1 -> 7.0.0 - Xetabase 2.2.1 -> 3.0.0
opencb · Oct 14, 2024 · 6aeead1 · 6aeead1
2 parents b663e0f + 454c24a
commit 6aeead1
Show file tree

Hide file tree

Showing 20 changed files with 221 additions and 250 deletions.
diff --git a/cellbase-app/app/cloud/docker/cellbase-builder/Dockerfile b/cellbase-app/app/cloud/docker/cellbase-builder/Dockerfile
@@ -28,4 +28,4 @@ RUN cd /opt/ensembl && \
     git clone https://github.com/Ensembl/ensembl-compara.git && \
     git clone https://github.com/Ensembl/ensembl-io.git
 
-ENV PERL5LIB=$PERL5LIB:/opt/ensembl/bioperl-live:/opt/ensembl/ensembl/modules:/opt/ensembl/ensembl-variation/modules:/opt/ensembl/ensembl-funcgen/modules:/opt/ensembl/ensembl-compara/modules:/opt/ensembl/lib/perl/5.18.2:/opt/cellbase/scripts/ensembl-scripts
+ENV PERL5LIB=$PERL5LIB:/opt/ensembl/bioperl-live:/opt/ensembl/ensembl/modules:/opt/ensembl/ensembl-variation/modules:/opt/ensembl/ensembl-funcgen/modules:/opt/ensembl/ensembl-compara/modules:/opt/ensembl/lib/perl/5.18.2:/opt/cellbase
diff --git a/cellbase-app/app/scripts/ensembl-scripts/DB_CONFIG.pm b/cellbase-app/app/scripts/ensembl-scripts/DB_CONFIG.pm
@@ -134,10 +134,10 @@ our $ENSEMBL_GENOMES_PORT = "4157";
 our $ENSEMBL_GENOMES_USER = "anonymous";
 
 ## Vertebrates
-our $HOMO_SAPIENS_CORE = "homo_sapiens_core_110_38";
-our $HOMO_SAPIENS_VARIATION = "homo_sapiens_variation_110_38";
-our $HOMO_SAPIENS_FUNCTIONAL = "homo_sapiens_funcgen_110_38";
-our $HOMO_SAPIENS_COMPARA = "homo_sapiens_compara_110_38";
+our $HOMO_SAPIENS_CORE = "homo_sapiens_core_104_38";
+our $HOMO_SAPIENS_VARIATION = "homo_sapiens_variation_104_38";
+our $HOMO_SAPIENS_FUNCTIONAL = "homo_sapiens_funcgen_104_38";
+our $HOMO_SAPIENS_COMPARA = "homo_sapiens_compara_104_38";
 #our $HOMO_SAPIENS_CORE = "homo_sapiens_core_78_38";
 #our $HOMO_SAPIENS_VARIATION = "homo_sapiens_variation_78_38";
 #our $HOMO_SAPIENS_FUNCTIONAL = "homo_sapiens_funcgen_78_38";

diff --git a/cellbase-core/src/main/java/org/opencb/cellbase/core/config/DownloadProperties.java b/cellbase-core/src/main/java/org/opencb/cellbase/core/config/DownloadProperties.java
@@ -26,7 +26,6 @@ public class DownloadProperties {
     private EnsemblProperties ensembl;
     private EnsemblProperties ensemblGenomes;
     private URLProperties hgnc;
-    private URLProperties cancerHotspot;
     private URLProperties refSeq;
     private URLProperties refSeqFasta;
     private URLProperties refSeqProteinFasta;
@@ -72,7 +71,6 @@ public class DownloadProperties {
     private URLProperties hpoObo;
     private URLProperties goObo;
     private URLProperties doidObo;
-    private URLProperties mondoObo;
     private URLProperties goAnnotation;
     private URLProperties revel;
     private URLProperties pubmed;
@@ -529,24 +527,6 @@ public DownloadProperties setHgnc(URLProperties hgnc) {
         return this;
     }
 
-    public URLProperties getCancerHotspot() {
-        return cancerHotspot;
-    }
-
-    public DownloadProperties setCancerHotspot(URLProperties cancerHotspot) {
-        this.cancerHotspot = cancerHotspot;
-        return this;
-    }
-
-    public URLProperties getMondoObo() {
-        return mondoObo;
-    }
-
-    public DownloadProperties setMondoObo(URLProperties mondoObo) {
-        this.mondoObo = mondoObo;
-        return this;
-    }
-
     public static class EnsemblProperties {
 
         private DatabaseCredentials database;

diff --git a/cellbase-core/src/main/resources/configuration.yml b/cellbase-core/src/main/resources/configuration.yml
@@ -62,11 +62,7 @@ download:
     url:
       host: ftp://ftp.ensemblgenomes.org/pub
   hgnc:
-    host: https://ftp.ebi.ac.uk/pub/databases/genenames/hgnc/archive/monthly/tsv/hgnc_complete_set_2023-11-01.txt
-    version: 2023-11-01
-  cancerHotspot:
-    host: https://www.cancerhotspots.org/files/hotspots_v2.xls
-    version: "v2"
+    host: https://ftp.ebi.ac.uk/pub/databases/genenames/hgnc/archive/monthly/tsv/hgnc_complete_set_2022-01-01.txt
   refSeq:
     host: https://ftp.ncbi.nih.gov/refseq/H_sapiens/annotation/GRCh38_latest/refseq_identifiers/GRCh38_latest_genomic.gtf.gz
   refSeqFasta:
@@ -77,15 +73,12 @@ download:
     host: https://ftp.ncbi.nih.gov/refseq/H_sapiens/annotation/GRCh38_latest/refseq_identifiers/GRCh38_latest_rna.fna.gz
   maneSelect:
 #    host: https://ftp.ncbi.nlm.nih.gov/refseq/MANE/MANE_human/release_0.93/MANE.GRCh38.v0.93.summary.txt.gz
-#    host: https://ftp.ncbi.nlm.nih.gov/refseq/MANE/MANE_human/release_1.0/MANE.GRCh38.v1.0.summary.txt.gz
-    host: https://ftp.ncbi.nlm.nih.gov/refseq/MANE/MANE_human/release_1.1/MANE.GRCh38.v1.1.summary.txt.gz
-    version: "1.1"
+    host: https://ftp.ncbi.nlm.nih.gov/refseq/MANE/MANE_human/release_1.0/MANE.GRCh38.v1.0.summary.txt.gz
+    version: 0.93
   lrg:
     host: http://ftp.ebi.ac.uk/pub/databases/lrgex/list_LRGs_transcripts_xrefs.txt
-    version: "2021-03-30"
   geneUniprotXref:
     host: http://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/idmapping/by_organism/
-    version: "2023-11-08"
   geneExpressionAtlas:
     host: ftp://ftp.ebi.ac.uk/pub/databases/microarray/data/gxa/allgenes_updown_in_organism_part_2.0.14.tab.gz
   mirbase:
@@ -95,49 +88,45 @@ download:
   targetScan:
     host: http://hgdownload.cse.ucsc.edu/goldenPath/
   miRTarBase:
-    host: https://mirtarbase.cuhk.edu.cn/~miRTarBase/miRTarBase_2022/cache/download/9.0/hsa_MTI.xlsx
-    version: "9.0"
-
-  ## Protein Data
+    host: https://mirtarbase.cuhk.edu.cn/~miRTarBase/miRTarBase_2022/cache/download/8.0/hsa_MTI.xlsx
   uniprot:
-    host: https://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.xml.gz
-    version: "2023-11-08"
+    host: ftp://ftp.uniprot.org/pub/databases/uniprot/current_release/knowledgebase/complete/uniprot_sprot.xml.gz
   uniprotRelNotes:
-    host: https://ftp.uniprot.org/pub/databases/uniprot/relnotes.txt
-    version: "2023-11-08"
+    host: ftp://ftp.uniprot.org/pub/databases/uniprot/relnotes.txt
+  intact:
+    host: ftp://ftp.ebi.ac.uk/pub/databases/intact/current/psimitab/intact.txt
   interpro:
-    host: https://ftp.ebi.ac.uk/pub/databases/interpro/current_release/protein2ipr.dat.gz
-    version: "2023-11-08"
+    host: ftp://ftp.ebi.ac.uk/pub/databases/interpro/current/protein2ipr.dat.gz
   interproRelNotes:
-    host: https://ftp.ebi.ac.uk/pub/databases/interpro/current_release/release_notes.txt
-  intact:
-    host: https://ftp.ebi.ac.uk/pub/databases/intact/current/psimitab/intact.txt
-    version: "2023-10-07"
-
-  ## Conservation Scores
+    host: ftp://ftp.ebi.ac.uk/pub/databases/interpro/current/release_notes.txt
   conservation:
     host: https://hgdownload.cse.ucsc.edu/goldenPath/
-    version: "2022-08-30"
   gerp:
-    host: http://ftp.ensembl.org/pub/release-110/compara/conservation_scores/91_mammals.gerp_conservation_score/gerp_conservation_scores.homo_sapiens.GRCh38.bw
-    version: "2023-05-17"
+    host: http://ftp.ensembl.org/pub/release-104/compara/conservation_scores/90_mammals.gerp_conservation_score/gerp_conservation_scores.homo_sapiens.GRCh38.bw
   clinvar:
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2021-07.xml.gz
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2022-02.xml.gz
+<<<<<<< HEAD
+    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2022-11.xml.gz
+  clinvarVariation:
+#    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2021-07.xml.gz
+#    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2022-02.xml.gz
+    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2022-11.xml.gz
+=======
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2022-11.xml.gz
-    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/ClinVarFullRelease_2023-12.xml.gz
-    version: "2023-12-01"
+    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/RCV_xml_old_format/ClinVarFullRelease_2024-05.xml.gz
+    version: 2024-05
   clinvarVariation:
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2021-07.xml.gz
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2022-02.xml.gz
 #    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2022-11.xml.gz
-    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/clinvar_variation/ClinVarVariationRelease_2023-12.xml.gz
+    host: https://ftp.ncbi.nlm.nih.gov/pub/clinvar/xml/VCV_xml_old_format/ClinVarVariationRelease_2024-05.xml.gz
+    version: 2024-05
+>>>>>>> release-6.2.x
   clinvarSummary:
     host: http://ftp.ncbi.nlm.nih.gov/pub/clinvar/tab_delimited/variant_summary.txt.gz
-    version: "2023-12-01"
   clinvarVariationAllele:
     host: http://ftp.ncbi.nlm.nih.gov/pub/clinvar/tab_delimited/variation_allele.txt.gz
-    version: "2023-12-01"
   clinvarEfoTerms:
     host: ftp://ftp.ebi.ac.uk/pub/databases/eva/ClinVar/2015/ClinVar_Traits_EFO_Names_260615.csv
   dbSNP:
@@ -158,43 +147,37 @@ download:
   genomicSuperDups:
     host: http://hgdownload.cse.ucsc.edu/goldenPath
   gwasCatalog:
-#    host: http://resources.opencb.org/opencb/cellbase/data/gwas/gwas_catalog_v1.0.2-associations_e106_r2022-05-17.tsv
-#    version: "1.0.2 associations_e106_r2022-05-17"
-    host: ftp://ftp.ebi.ac.uk/pub/databases/gwas/releases/2023/12/21/gwas-catalog-associations.tsv
-    version: "23-12-21"
+<<<<<<< HEAD
+    host: http://resources.opencb.org/opencb/cellbase/data/gwas/gwas_catalog_v1.0.2-associations_e106_r2022-05-17.tsv
+    version: "1.0.2 associations_e106_r2022-05-17"
+=======
+    #host: http://resources.opencb.org/opencb/cellbase/data/gwas/gwas_catalog_v1.0.2-associations_e106_r2022-05-17.tsv
+    host: "https://ftp.ebi.ac.uk/pub/databases/gwas/releases/2024/05/20/gwas-catalog-associations_ontology-annotated.tsv"
+    #version: "1.0.2 associations_e106_r2022-05-17"
+    version: "2024-05-20"
+>>>>>>> release-6.2.x
   hpo:
-    ## Downlaod manually from here now:  https://hpo.jax.org/app/data/annotations
     host: https://ci.monarchinitiative.org/view/hpo/job/hpo.annotations/lastSuccessfulBuild/artifact/rare-diseases/util/annotation/phenotype_to_genes.txt
   disgenet:
     host: https://www.disgenet.org/static/disgenet_ap1/files/downloads
     files:
       - all_gene_disease_associations.tsv.gz
       - readme.txt
   dgidb:
-    host: https://old.dgidb.org/data/monthly_tsvs/2022-Feb/interactions.tsv
-    version: "2022-02-01"
+    host: https://dgidb.org/data/monthly_tsvs/2021-Jan/interactions.tsv
   cadd:
-    ## Nacho: Move to https://krishna.gs.washington.edu/download/CADD/v1.7-pre/GRCh38/whole_genome_SNVs.tsv.gz ASAP!
-#    host: https://krishna.gs.washington.edu/download/CADD/v1.6/GRCh38/whole_genome_SNVs.tsv.gz
-    host: https://krishna.gs.washington.edu/download/CADD/v1.7-pre/GRCh38/whole_genome_SNVs.tsv.gz
-    version: "1.7-pre"
+    host: https://krishna.gs.washington.edu/download/CADD/v1.6/GRCh38/whole_genome_SNVs.tsv.gz
   reactome:
     host: http://www.reactome.org/download/current/biopax.zip
   gnomadConstraints:
     host: https://storage.googleapis.com/gcp-public-data--gnomad/release/2.1.1/constraint/gnomad.v2.1.1.lof_metrics.by_transcript.txt.bgz
-    version: "2.1.1"
+    version: 2.1.1
   hpoObo:
     host: http://purl.obolibrary.org/obo/hp.obo
-    version: "2023-12-01"
   goObo:
     host: http://purl.obolibrary.org/obo/go/go-basic.obo
-    version: "2023-12-01"
   doidObo:
     host: http://purl.obolibrary.org/obo/doid.obo
-    version: "2023-12-01"
-  mondoObo:
-    host: http://purl.obolibrary.org/obo/mondo.obo
-    version: "2023-12-01"
   goAnnotation:
     host: http://geneontology.org/gene-associations/goa_human.gaf.gz
   revel:
@@ -221,7 +204,7 @@ species:
     - id: hsapiens
       scientificName: Homo sapiens
       assemblies:
-        - ensemblVersion: '110_38'
+        - ensemblVersion: '104_38'
           name: GRCh38
         - ensemblVersion: '82_37'
           name: GRCh37

diff --git a/cellbase-lib/pom.xml b/cellbase-lib/pom.xml
@@ -137,10 +137,10 @@
             <groupId>com.github.samtools</groupId>
             <artifactId>htsjdk</artifactId>
         </dependency>
-<!--        <dependency>
+        <dependency>
             <groupId>org.codehaus.jackson</groupId>
             <artifactId>jackson-mapper-asl</artifactId>
-        </dependency>-->
+        </dependency>
         <dependency>
             <groupId>io.jsonwebtoken</groupId>
             <artifactId>jjwt-api</artifactId>

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java
@@ -56,9 +56,9 @@ public class EtlCommons {
     public static final String PHARMGKB_VERSION_FILENAME = "pharmgkbVersion.json";
 
     public static final String CLINICAL_VARIANTS_FOLDER = "clinicalVariant";
-    public static final String CLINVAR_VERSION = "2022.11";
-    public static final String CLINVAR_DATE = "2022-11";
-    public static final String CLINVAR_XML_FILE = "ClinVarFullRelease_2022-11.xml.gz";
+    public static final String CLINVAR_VERSION = "2024-05";
+    public static final String CLINVAR_DATE = "2024-05";
+    public static final String CLINVAR_XML_FILE = "ClinVarFullRelease_2024-05.xml.gz";
     public static final String CLINVAR_EFO_FILE = "ClinVar_Traits_EFO_Names.csv";
     public static final String CLINVAR_SUMMARY_FILE = "variant_summary.txt.gz";
     public static final String CLINVAR_VARIATION_ALLELE_FILE = "variation_allele.txt.gz";
@@ -77,7 +77,6 @@ public class EtlCommons {
     public static final String HPO_FILE = "hp.obo";
     public static final String GO_FILE = "go-basic.obo";
     public static final String DOID_FILE = "doid.obo";
-    public static final String MONDO_FILE = "mondo.obo";
     public static final String PFM_DATA = "regulatory_pfm";
 
     // Build specific data options

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/GeneBuilder.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/GeneBuilder.java
@@ -90,8 +90,8 @@ public GeneBuilder(Path geneDirectoryPath, Path genomeSequenceFastaFile, Species
                        boolean flexibleGTFParsing, CellBaseSerializer serializer) throws CellBaseException {
         this(null, geneDirectoryPath.resolve("description.txt"),
                 geneDirectoryPath.resolve("xrefs.txt"),
-                geneDirectoryPath.resolve("hgnc_complete_set_2023-11-01.txt"),
-                geneDirectoryPath.resolve("MANE.GRCh38.v1.1.summary.txt.gz"),
+                geneDirectoryPath.resolve("hgnc_complete_set_2022-01-01.txt"),
+                geneDirectoryPath.resolve("MANE.GRCh38.v1.0.summary.txt.gz"),
                 geneDirectoryPath.resolve("list_LRGs_transcripts_xrefs.txt"),
                 geneDirectoryPath.resolve("idmapping_selected.tab.gz"),
                 geneDirectoryPath.getParent().resolve("regulation/motif_features.gff.gz"),

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/OntologyBuilder.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/OntologyBuilder.java
@@ -32,14 +32,12 @@ public class OntologyBuilder extends CellBaseBuilder {
     private Path hpoFile;
     private Path goFile;
     private Path doidFile;
-    private Path mondoFile;
 
     public OntologyBuilder(Path oboDirectoryPath, CellBaseSerializer serializer) {
         super(serializer);
         hpoFile = oboDirectoryPath.resolve(EtlCommons.HPO_FILE);
         goFile = oboDirectoryPath.resolve(EtlCommons.GO_FILE);
         doidFile = oboDirectoryPath.resolve(EtlCommons.DOID_FILE);
-        mondoFile = oboDirectoryPath.resolve(EtlCommons.MONDO_FILE);
     }
 
     @Override
@@ -66,13 +64,6 @@ public void parse() throws Exception {
             serializer.serialize(term);
         }
 
-        bufferedReader = FileUtils.newBufferedReader(mondoFile);
-        terms = parser.parseOBO(bufferedReader, "Mondo Ontology");
-        for (OntologyTerm term : terms) {
-            term.setSource("MONDO");
-            serializer.serialize(term);
-        }
-
         serializer.close();
     }
 }